اثر محاسبه ذهنی همزمان بر روی لکنت زبان، استنشاق و زمان بندی سخنرانی

The aim of the present investigation was to test the assumption that concurrent cognitive processes interfere with the fluent execution of speech movements. During speech production, information related to subsequent speech is retrieved from short-term memory and translated into a phonetic code while previously planned portions of speech are being produced Ferreira 1991 and Ferreira 1993. Therefore it is assumed that the processing load continuously fluctuates during speech production Power 1985 and Ford & Holmes 1978 and that these fluctuations are correlated with variations in the frequency of stuttering. In the present experiment, a dual-task paradigm was used to investigate whether higher-level cognitive processing can induce speech disfluencies and how persons who stutter differ from persons who do not stutter with respect to frequency of inhalation and speech rate (Bosshardt, 1997). Under dual-task conditions, an automatic word repetition task had to be performed concurrently with mental addition as a secondary task. In the oral word-repetition task, participants were required to continuously repeat a sequence of three words until the experimenter stopped them after the 10th repetition. Speaking under dual-task conditions has already been studied in stuttering research Arends et al. 1988, Brutten & Trotter 1985, Brutten & Trotter 1986, Kamhi and McOsker 1982, Mallard & Webb 1980, Sussman 1982 and Thompson 1985. Mallard and Webb found that irregular changes of the room illumination and presentation of irrelevant speech noise did not significantly affect speech fluency. Similarly, Kamhi and McOsker (1982) found that a nonattention-demanding gross-motor task also had no effect on speech fluency. As an attention-demanding task, pursuit tracking requires continuous sensory-motor coordination. Thompson (1985) did not find that tracking as a secondary task produced significant changes in stuttering rate; however, with more complex speech tasks, Arends et al. (1988) found that pursuit tracking affects speech fluency. In particular, they found that in spontaneous speech production the number and duration of disfluencies was significantly smaller when this task had to be performed concurrently with tracking. From the theory of multiple resources as proposed by Wickens (1984)(see also Wieland-Eckelmann, 1992), it can be deduced that the degree to which speech fluency is affected by a secondary task depends on the extent to which both tasks draw on the same processing resources. In the present investigation overt word repetition and mental calculation were chosen as dual tasks because on the background of the existing evidence Baddeley 1997, Baddeley, Eldridge, & Lewis 1981 and Gathercole & Baddeley 1993 it can be assumed that both tasks demand resources from the central executive and phonological loop systems. Presumably, continuous word repetition can essentially be seen as a product of the phonological loop and of the articulatory systems with minimal demands on the central executive, whereas conversely the mental addition task is largely based on central executive processes (Hitch, 1978). In mental calculation, the phonological system is only involved insofar as provisional results are stored in it. The central executive is a limited-capacity system which is responsible for the control of mental processes and short-term retention with one strategy for the latter being use of the phonological loop for the storage of provisional results. The empirical and theoretical evidence reviewed so far leads to two opposing expectations. Insofar as word repetition and mental calculation draw on identical processing resources, it can be assumed that speech disfluencies increase under dual-task conditions. On the other hand, the results of Arends et al. (1988) show that under dual-task conditions speech disfluencies may even decrease. This result can be taken as an indication that under dual-task conditions additional processing resources can come into effect which were not equally effective under single-task speaking. This result can then be interpreted as a demonstration of a form of “controlled” or “supervised” speaking which reduces the amount of disfluency and which is more effective when persons are speaking under dual-task conditions (Bosshardt, 1997). This supervised form of speaking can be seen as a change in the control mode for speech as suggested by van Lieshout, Hulstijn, and Peters 1996a and van Lieshout, Hulstijn, and Peters 1996b. It can be assumed that a change in the control mode does not only affect speech fluency but speech rate as well. It is conceivable that a supervised form of speaking in which the articulatory movements are continuously controlled by feedback needs more time than automatic speaking. Therefore, the present investigation also examined how concurrent mental addition affects word duration and how changes in word duration are related to fluency. Another strategy to reduce the processing load imposed by the dual-task conditions would be to increase the frequency of inhalations. Inhalations can be used as delays during which mental calculations can be performed. Therefore, another research question to be addressed in the present experiment was whether the effects of dual task on stuttering are moderated by breathing. The mental calculation task differs in two important ways from the secondary tasks that were used in earlier experiments. The first difference is related to the fact that in the present experiment the mental calculation task was presented at a particular moment during word repetition. This makes it possible to investigate whether speech fluency is already increased or decreased when a secondary task is only expected but not yet actually presented. Another difference between the mental calculation task in the present experiment and the secondary tasks of earlier investigations is that it does not require an overt motor reaction. It is thus possible to determine whether concurrent cognitive activities affect the fluency of highly automatized word repetition. The effects of physical perturbations on oscillatory systems were frequently found to be different when applied in different phases of the system (Glass & Mackey, 1988). It is conceivable that the effects of cognitive perturbation as used in the present experiment could also depend on phase relations. In the present experiment three words were continuously repeated with the respiratory cycle superimposed. Therefore, care was taken to synchronize the application of the perturbation with the word repetition and respiratory cycles. The mental calculation task was always presented with the last word of the third repetition thus guaranteeing that the secondary task was presented within the exhalation phase of a respiratory cycle. Lively, Pisoni, van Summers, & Bernacki (1993) made acoustic measurements of test sentences produced under simultaneous visual tracking and under control conditions. Only persons who do not stutter participated in this investigation. Most importantly, the effects of dual-task manipulations on speech were individually determined for every speaker. Lively et al. (1993) found, for example, that under dual-task conditions four of the five participants had significantly shorter phrase durations but one speaker showed the opposite pattern. Similar individual differences were found in other measures, like speech amplitude and amplitude variation. The results of this study show that it may be misleading to base conclusions on aggregated group data alone. In the present study, not only aggregated data but also the data of individual subjects were statistically analyzed. The aim of the present study was to investigate how the fluency of word repetition is affected when the processing space of the central executive and loop system is reduced by a concurrent mental calculation task. More specifically, the following research questions were addressed: 1) How is the fluency of the word repetitions, word duration, and breathing activity influenced when mental calculation has to be performed concurrently? 2) Are there individual differences with respect to the way in which mental calculation changes the fluency of word repetitions, word duration, or breathing? 3) Are the effects of dual task on stuttering moderated by breathing and word duration? Method Subjects Nine adults who stutter (mean age = 37.2 years; SD = 11.2) and 10 who do not stutter (mean age = 23.1 years; SD = 4.3) participated in the investigation. The participants who do not stutter were undergraduate students of psychology who participated in the experiment to fulfill their course requirements. The persons who stutter were recruited from self-help groups. As determined in oral reading of a newspaper text of 200 words these people stuttered (silent prolongations, prolongations, and repetitions of sounds, syllables, and one-syllable words) on the average on 6.2% (SD = 5.2) of the words. In a subsequent free report of this text, the stuttering rates were 10.7% (SD = 9.9). In no case was the duration of the three longest blocks longer than 1 second. Table 1 presents the individual data for all participants. Table 1. Description of the Personal Characteristics of Individual Participants Participant identification Sex Age (years) Mental calculation Percentage of words stuttered (%) % Correct (N = 30) Reading Monologuea N01 F 29 87 — — N02 F 29 83 — — N03 F 20 97 — — N04 F 20 77 — — N05 F 30 87 — — N06 M 20 93 — — N07 M 20 87 — — N08 M 21 90 — — N09 M 21 97 — — N10 M 21 97 — — S01 M 27 80 6.5 6.5 (139) S02 M 23 90 2.5 12.7 (228) S03 M 52 97 11.5 20.6 (165) S04 M 48 90 2.0 7.5 (187) S05 F 33 93 9.0 4.0 (198) S06 M 45 87 1.5 10.0 (200) S07 M 49 93 0.5 2.5 (201) S08 M 27 87 12.0 30.9 (123) S09 M 31 100 4.5 11.1 (81) a Number of words in parentheses. Table options The two groups of participants were matched for single-task mental calculation performance. For every participant, the percentage of correct solutions in single-task mental calculation was determined for every session. The groups were matched on the basis of percentages averaged over sessions. Three of the persons who stutter had the lowest mental calculation performance (73% and 70%) and had to be excluded from the following analyses because not one person who did not stutter had a comparable performance. Those who stuttered had 90.8% correct solutions (SD = 6.0; n = 9) as compared to 89.5% (SD = 6.7; n = 10) for those who did not stutter. The difference between the means of the two groups was not significant (t(17) = 0.48; ns), and the variances were homogeneous . Material The words for the repetition task were drawn from a pool of 10 three-syllabic compound nouns: Bohnentopf, Blattspinat, Buttermilch, Reissalat, Suppenkraut, Knoblauchbrot, Kokosnuss, Sauerkraut, Traubensaft, Weizengrieß (translation: pot of beans, leafy spinach, buttermilk, rice salad, soupstock seasoning, bread with garlic, coconut, pickled cabbage, grapejuice, semolina). These nouns are all common words from the semantic field foodstuffs, have three syllables, are pronounced with primary stress on the first syllable, and begin or end with plosives or fricatives. From this pool, 30 sequences of three words were pseudorandomly constructed with the restriction that at every position each word appeared at least in one and at most in six sequences. Ten sequences were used at each of the three sessions. A total number of 30 addition tasks was constructed. For these tasks the first addends were between 21 and 44, the second between 11 and 18, and the third between 2 and 9. The sum of the first two addends required no carrying, but addition of the third always did. Procedure The experiment was run in three sessions on three days. The single and dual tasks in every session consisted of 10 trials. Each session began with 10 trials of mental addition as a single task and was followed by 10 word-repetition trials as a single task. The third and final part in every session always consisted of 10 dual-task trials. Every trial was started by the participants operating a push button causing a computer program to present three words on the monitor. As soon as the experimenter operated one of his buttons, the same three words plus three numbers were exposed for three seconds. The experimenter exposed the numbers at the end of the third repetition of the words when the stressed syllable of the third word was being pronounced. After three seconds, the numbers were automatically cleared while the words remained visible. Within a session the same material was presented visually in all single- and dual-task conditions. In the mental addition part of the investigation, participants were instructed to disregard the words, calculate the sum, and produce the result as fast as possible. In the word repetition part, the same material was presented again with the instruction to repeat the three words continuously as fast as possible and disregard the numbers. After 10 vocal repetitions of the words the experimenter indicated the end of the trial and cleared the computer screen. In the dual-task part of the investigation, participants were instructed to mentally calculate while concurrently repeating the word list. As before, they were instructed to repeat the words continuously as fast as possible until the experimenter announced the end of the trial. Thus, under dual-task conditions participants performed the first three word repetitions with the expectation that they will be required to mentally add the summands as soon as the summands were exposed. In contrast, the word repetitions after the third one had to be produced concurrently with the secondary mental addition task. After the 10th repetition the experimenter announced the end of the trial, and the subjects produced the result of their mental addition. Dependent Variables Stuttering Rate All repetitions, exchanges of sounds or syllables, together with lengthening of vowels and consonants, and all indications of tension (intensity and duration) were counted as stuttering. For every participant, repetition, and trial the percentage of syllables stuttered was determined. Inhalation Rate All audible signs of inhalation were registered and for every repetition it was determined how many inhalations occurred during the production of nine syllables belonging to one repetition of three words. An inhalation which immediately preceded a particular syllable was counted as an inhalation related to the production of this syllable. For every participant, repetition, and trial the percentage of syllables at which an inhalation occurred was determined. Length-of-Breath Group The length-of-breath group was defined as the number of words produced on one exhalation. For every participant and trial, the number of words which were produced on the first three exhalations was counted. Because five participants (two persons who stutter and three persons who do not stutter) repeated the words within less than three breath groups they had missing values on some conditions. In order to obtain a more complete set of data, the scores for every participant were averaged over sessions.1 These averaged data were complete for all except one person who does not stutter and who repeated all words in all trials and sessions within a single-breath group. Word Duration Word duration was determined in milliseconds from a spectrogram (Computerized Speech Lab, Kay) with 146 Hz bandwidth (10-kHz sampling rate). For words beginning with a stop consonant, the onset of the acoustic energy in the initial burst was used as word onset. For words beginning with /s/, /r/, and /w/, the onset of the acoustic energy was taken as word onset. The last vertical striation in the second or higher formants related to the last vowel of the word was taken as the word offset. For “Buttermilch” (buttermilk), the end of the /l/ was taken as the end of the word. Because these measurements were rather time-consuming they could only be performed for the word repetitions of the first session. Word duration was only determined for words which were spoken without any indication of disfluency. Reliability The criteria for scoring were defined by two raters using the verbal repetitions of one participant. Then these raters independently scored the verbal repetitions of four persons (N01, N10, S04, S192). Interrater agreement was determined for a total of 16,128 syllables. The raw percentages of syllables that were scored identically by both raters were 99.3% for stuttering and 99.7% for inhalation. After correcting for chance (Scott, 1955) the agreement between the raters was 79.8% for stuttering and 95.2% for inhalation. All scores used in this analysis were determined by one of the raters who in case of doubt discussed the scoring with the author. For a total of 1080 words (spoken by one person who stutters and one person who does not stutter), duration was independently measured by two raters. The interrater correlation was highly significant (r = .993; t(1078) = 268.05; p < .000). The intercept of the regression line between the two measures deviated only −2 ms from zero, and its slope was very close to one (0.994). Design In the single-subject analyses, stuttering rate and inhalation were treated as separate dependent variables with trial as a random factor nested within session (three). Thus, session was treated as a between-trials factor and repetition (1–10) and task (single vs. dual) as within-trial repeated measurement factors. Because word duration was only measured for the first session, the design for this variable consisted of the within-trial factors repetition and task. In addition to these single-subject analyses, between-group comparisons were also calculated. For these latter analyses, the scores of every participant were averaged over the 10 trials of one session. For repeated measurement factors with more than one degree of freedom, a Greenhouse-Geisser correction is available, and the statistical significance of these factors was always determined after correcting their degrees of freedom with ϵ; only the corrected error probabilities will be reported. Results The present investigation is focused on the effects of mental calculation on speaking. But within a dual-task paradigm, speech fluency and timing data have to be interpreted in relation to the mental calculation performance. Therefore, the results of the mental calculation performance are presented first. Mental Calculation Performance The two groups of participants were matched with respect to their single-task mental calculation performance. The percentages of correct solutions for mental calculation form a three-factorial design with group as a between-subjects factor, session and task as repeated-measurement factors. The factor group (F(1,17) = 0.46; MSE = 486.7; ns) and the interactions of the group factor with task (F(1,17) = 0.37; MSE = 147.6; ns), with session (F(1,17) = 0.66; MSE = 94.9; ϵ = 0.99; ns) and with task × session (F(1,17) = 2.25; MSE = 98.7; ϵ = 0.99; ns) did not significantly influence mental calculation performance. These results imply that the two groups were comparable in how the mental calculation performance was influenced by the factors task, session, and task × session. But the mental calculation performance was significantly influenced by session (F(4,17) = 8.21; MSE = 98.7; ϵ = 0.99; p < .001), task (F(1,17) = 20.35; MSE = 147.6; ϵ = 0.99; p < .000), and session × task (F(1,17) = 3.32; MSE = 98.7; ϵ = 0.99; p < .05). The percentage of correct solutions was significantly higher under single- as compared to dual-task performance (single task: M = 90.0%; SD = 6.2; dual task: M = 79.6%; SD = 12.9), and this difference was more pronounced in the first two sessions than in the last session (session 1: 88% vs. 72%; session 2: 93% vs. 85%; session 3: 88% vs. 83%). The task effect was only significant for the first (F(1,17) = 23.48; MSE = 115.9; p < .000) and second (F(1,17) = 5.91; MSE = 104.8; p < .03), but not for the last session (F(1,17) = 2.58; MSE = 125.0; ns). This result is taken as an indication that the difficulty of the mental calculation performance was reduced over sessions so that in the last session mental calculation performance was not significantly reduced under dual- as compared to single-task conditions. Stuttering Rate In the between-group analyses stuttering rate was significantly influenced by the interaction task × repetition (F(9,153) = 8.35; MSE = 4.26; p < .001; ϵ = 0.25). Figure 1 shows that relative to the stuttering rates under single-task conditions, stuttering rates under dual-task conditions decreased in repetitions 2 and 3 (F(1,17) = 33.00; MSE = 2.2; p < .000 and F(1,17) = 14.21; MSE = 2.4; p < .01) and increased in repetitions 4 and 5 (F(1,17) = 11.19; MSE = 13.4; p < .01 and F(1,17) = 5.5; MSE = 6.5; p < .05). The task effect was not significant in the other repetitions. The three-way interaction between group, repetition, and task failed to be significant (F(9,153) = 1.09; MSE = 4.26; n.s.). Of the other main effects and interactions, only the repetition effect was also significant (F(9,153) = 8.17; MSE = 4.73; p < .000, ϵ = 0.28). Thus, the between-group comparisons showed that speech fluency of the two groups was influenced in a comparable way by the secondary task: immediately before the presentation of the mental calculation task (i.e., in repetitions 2 and 3) their stuttering rate significantly decreased, and it increased during repetitions 4 and 5 when mental calculation is actually being performed. Full-size image (12 K) Figure 1. Stuttering rates for the two groups of individuals under single- and dual-task conditions Figure options The factor session (F(2,34) = 0.47, MSE = 20.08; ns) and the interactions with this factor did not significantly influence the stuttering rate (all error probabilities under H0 were .22 or greater). In contrast to the mental calculation performance, the between-subject analyses of stuttering rate did not provide any evidence that the stuttering rate was significantly influenced by session. Separate analyses of variance were conducted for every participant with session as between-trial factor and task and repetition as within-trial repeated measurement factors. The simple effects of task at each repetition were only calculated when the effect of task and/or the interaction task × repetition were significant for this participant. In Figures 2a and 2b the results of these tests are presented for every participant. The participants in this and subsequent figures cited in this report (4a, 4b, 6a, and 6b) are sorted with respect to the difference between their stuttering rates in repetition 4 with the participants showing the largest difference on the left and those with the smallest or negative differences on the right. Full-size image (32 K) Full-size image (31 K) Figure 2. Change of the stuttering rates from single- to dual-task conditions: (a) Persons who do not stutter. (b) Persons who stutter. Persons are sorted from left to right according to their difference score in repetition 4. Asterisk represents significant changes (see text for explanation) Figure options These single-subject analyses of stuttering rate revealed that under dual-task conditions three persons who stutter reacted to the mental calculation task in repetition 4 with an extremely large increase in stuttering rate (see S01, S04, S08 in Figure 2b) while others reacted in a way that was comparable to that of persons who do not stutter. This result indicates that mental calculation as a secondary task increases existing differences between the stuttering rates of persons who stutter. This assumption can be statistically evaluated by analyzing the interaction subjects × task calculated within each group and repetition (Table 2). In repetition 4, this variance component was significantly larger for the group of persons who stutter than the corresponding variance for persons who do not stutter. Table 2 shows that the mental addition task increases existing interindividual differences between the stuttering rates of persons who stutter to a significantly larger extent than that of persons who do not stutter. Some persons who stutter react to the dual-task condition with an increase in stuttering rate which is much higher than that of other persons who stutter and also than that of persons who do not stutter (see Figures 2a and 2b) while others even decrease their stuttering rate under dual-task conditions below the level characteristic of single task. However, it is remarkable that under dual-task conditions, significant decreases of stuttering rates were only found before the presentation of the mental calculation task (for person N09 in repetition 2) or after it (for N04 in repetition 9) but in no case in any of the repetitions between 4 and 8. Table 2. Mean Sum of Squares Subject × Task for the Two Groups of Persons and F-tests (see text for explanation) Repetition Stutterersa Nonstutterersb F(8, 9) Significance 1 6.36 3.19 1.99 ns 2 2.55 1.82 1.40 ns 3 3.92 1.13 3.46 p < .05 4 25.20 2.91 8.66 p < .01 5 10.20 2.93 3.48 p < .05 6 4.94 2.13 2.32 ns 7 1.01 1.29 0.78 ns 8 0.44 2.40 0.18 ns 9 4.44 2.21 2.01 ns 10 7.79 2.76 2.82 ns a df = 8. b df = 9. Table options Inhalation Rate The between-group analysis of inhalation rate showed a significant task by repetition interaction (F(9,153) = 5.44; MSE = 4.08; p < .001; ϵ = 0.40). This interaction results from the fact that under dual-task conditions the mental calculation task temporarily increased the inhalation rate in repetition 4 (Figure 3). There was no indication of a statistically reliable interaction between task, repetition, and group (F(9,153) = 0.68; MSE = 4.08, ns), but the Full-size image (12 K) Figure 3. Inhalation rates for the two groups of persons under single- and dual-task conditions Figure options inhalation rate of persons who stutter was significantly higher than that of persons who do not stutter (F(1,17) = 23.46; MSE = 94.20; p < .000) and there was a significant interaction between group and task (F(1,17) = 4.33; MSE = 6.97; p < .05). These results indicate that under dual-task conditions immediately after the presentation of the mental calculation task both groups reacted with a temporary increase in inhalation rate. After this temporary increase in repetition 4, the inhalation rate of persons who do not stutter returned to the base level characteristic for single-task repetitions. Therefore, over all repetitions the inhalation rate of this group was slightly higher in dual- as compared to single-task speech (1.5% vs. 1.8%). Persons who stutter showed a different pattern because after their fourth repetition their inhalation rate was reduced below the level characteristic for single-task speaking. Therefore, over all repetitions, the respiration rate of persons who stutter was lower in dual- than in single-task conditions (4.6% vs. 4.3%). This pattern of results suggests that the two groups of persons differ in how the additional air that is taken under dual-task conditions affects the following breath cycles. Individual analyses of inhalation rate showed that under dual-task conditions seven persons significantly increased their inhalation rate in repetition 4 (three persons who stutter and four who do not stutter). The other participants did not show significant changes of inhalation rate during repetition 4 (Figure 4). Thus, under dual-task conditions about half of the participants from both groups significantly increased their inhalation rate whereas the other half did not significantly change their breathing pattern. Full-size image (34 K) Full-size image (32 K) Figure 4. Change of the inhalation rates from single- to dual-task conditions: (a) Persons who do not stutter. (b) Persons who stutter. Persons are sorted in the same way as in Figure 2a and 2b. Asterisk represents significant changes (see text for explanation) Figure options Length-of-Breath Group The length of the first, second, and third breath groups were counted and averaged for every participant over trials and sessions. Because one person who does not stutter had fewer than three inhalations in all trials, this person’s data were lacking in one condition and had to be excluded from the analysis. Analysis of variance of these data showed significant effects of group (F(1,16) = 20.9; MSE = 17.26; p < .000), breath group (F(2,32) = 21.88; MSE = 20.33; p < .000, ϵ = .58), breath group by task (F(2,32) = 5.37; MSE = 2.50; p < 0.14; ϵ = .85), and group by breath group by task (F(2,32) = 5.84; MSE = 2.50; p < .01; ϵ = .85). On the average, persons who do not stutter produced significantly more words within one breath group (10.8 words) than persons who did stutter (7.2 words), and both groups produced more words within the first than within the following breath group (Table 3). For a more detailed analysis of the three-way interaction between group, breath group, and task, the simple effects of the factor task were calculated separately for every breath group and for the two groups of persons. For persons who stutter none of these three simple effects reached the criterion of statistical significance (F(1,16) = 0.00; MSE = 3.09; F(1,16) = 0.19; MSE = 2.28; F(1,16) = 0.53; MSE = 0.91). Thus, for persons who stutter none of the first three breath groups was significantly influenced by task. In contrast, persons who do not stutter showed significant task effects in the first and third breath groups (F(1,16) = 13.54; MSE = 3.09; p < .002; F(1,16) = 0.60; MSE = 2.28; ns; F(1,16) = 14.13; MSE = 0.91; p < .002; ϵ = .58). Within the first breath group persons who do not stutter produced about three words fewer when speaking under dual-task than when speaking under single-task conditions (Table 3). The length of the second breath group was not significantly influenced by task. For persons who do not stutter the third expiration began under dual-task conditions on the average after the 24th word and after the 27th word in single-task. Because a total of 30 words had to be produced, under dual-task conditions six words were left over for production within the last breath group whereas under single-task conditions only three words were left for production within the third breath group. For persons who do not stutter, the prolongation of the third breath group under dual-task conditions seems to be a consequence of the shortening of the first breath group. The present results are in good agreement with the observations about the relationship between stuttering and speech rate Logan & Conture 1995, Logan & Conture 1997 and Yaruss 1997. These studies demonstrated that in spontaneous speech, speech rate and fluency are not reliably related to one another. Although these results were obtained with children, it seems likely that they equally apply to adults. The present results suggest that reductions in speech rate make it possible to perform other attention-demanding cognitive processes concurrently with speaking. It was found that at the moment when concurrent mental activities begin (i.e., in repetition 4) both those who stutter and those who do not temporarily decreased their speech rate. With this as a backdrop, it is conceivable that for individuals who stutter prolonged speech does not only offer opportunities for motor learning processes, rather, the fluency-enhancing effects of reductions in speech rate (see among others Andrews, Craig, Feyer, Hoddinott, & Neilson 1983 and Peters and Guitar 1991) may be mediated by cognitive processes because slow speech makes it possible that cognitive processes necessary for the supervision of speech production can become effective (a similar proposal was made by Yaruss, 1997). Conclusions for Group Differences Inhalation rate was the only dependent variable which showed significant interactions between group and task factors. It can be seen in Figure 3 that the two groups increase inhalation rate in a comparable way in repetition 4, i.e., when under dual-task conditions mental calculation is actually being performed. But in the following repetitions the two groups show a different inhalation behavior. Persons who stutter tend to reduce the inhalation rate in the following repetitions below the single-task level, whereas persons who do not stutter return to the base rate. For all repetitions, those subjects who stutter decrease and those who do not increase their inhalation rates under dual task conditions. Because in the present study only frequency of inhalation was observed we can only speculate about the possible reasons for this difference. The most probable interpretation of this group difference is that after repetition 4, persons who stutter reduce their speech efforts, whereas persons who do not stutter did not show a comparable change. Apart from the fact that in repetition 4 the two groups showed comparable increases in word duration under dual-task conditions, persons who stutter had generally longer word durations (614 ms) than persons who do not stutter (466 ms). Because word duration was determined for the fluently spoken words only, the stutterers’ slower articulation rate may be an indication of nonperceptible disfluencies. But it is equally conceivable that this slower articulation is the result of slower cognitive processing (see also Baddeley, 1997, p. 54ff). Because the articulation rate of both groups temporarily slowed down when mental addition had to be performed concurrently, this latter interpretation seems to provide a more parsimonious interpretation of the present results. In single-task word repetition, the stuttering rates of persons who stutter and of persons who do not stutter were not significantly different (persons who stutter: M = 2.6%, SD = 3.2; persons who do not stutter: M = 1.4%; SD = 0.7; t(8.7) = 1.22; ns). The word repetition task of the present experiment can be performed in a highly automatic way and its production was highly practiced because the same sequence had to be repeated 10 times. In addition, all sequences were composed of words from a pool of 10 words. It is to be expected that under these conditions an adaptation effect can be observed, i.e., that stuttering rate declines over repetitions. For persons who stutter the linear trend indicating a decrease of stuttering rate over repetitions was significant (F(1,17) = 10.46; MSE = 2.48; p < .01) but not for persons who do not stutter (F(1,17) = 3.80; MSE = 2.48; ns). Under dual-task conditions the stuttering rates of the two groups of persons were also not significantly different (persons who stutter: M = 3.2%, SD = 3.9; persons who do not stutter: M = 1.3%, SD = 0.5; t(8.7) = 1.34; ns); however, analyses of individual data (Figure 2a,b) revealed that within the group of persons who stutter, the stuttering rates of one subgroup increased significantly under dual-task conditions to an extent that was not comparable to any of the persons who do not stutter. Speech fluency of another subgroup of persons who stutter was influenced by dual-task conditions to an extent that was comparable to that of persons who do not stutter. In other words, the group of persons who stutter was heterogenous with respect to the extent to which their fluency was influenced by concurrent mental activity. Under dual-task conditions, the between-subject variance of persons who stutter increases significantly in repetitions 3, 4, and 5 (Table 2). These results indicate that the actual performance of a secondary task as well as the impendency of such a task can induce a significant increase in the variance of stuttering rates. The variances presented in Table 2 suggest that the group of persons who stutter can be divided into two subgroups. The variance increase is largely due to a subgroup of persons showing an extremely high increase in stuttering rate under dual-task conditions (high-interference subgroup). The stuttering rate of the other subgroup of persons who stutter is increased under dual-task conditions to a relatively smaller extent, comparable to that of persons who do not stutter (Figure 2a,b). Only provisional interpretations can be suggested for these differences between the subgroups. One interpretation4 is related to the fact that in the present experimental paradigm only highly practiced word sequences had to be repeated. It is conceivable that subgroups of persons differ in the amount to which their speech movements are sufficiently adapted after the third repetition so that they are able to minimize stuttering (Max, Caruso, & Vandevenne, 1997). According to the drop in activation interpretation developed earlier this implies that highly practiced speech motor programs can be run without errors even when they are only weakly activated. In this view it is assumed that the adaptation effect makes the speech of the low-interference subgroup resistant to interference from concurrent mental load. Implications for the Development of Stuttering The distribution of stuttering at particular linguistic loci already suggested that moments of enhanced processing load are moments at which the probability of stuttering is enhanced (for reviews, see St. Louis 1979 and Bernstein Ratner 1997). Because of the correlational character of these observations they are open to many alternative interpretations. The present experiment clearly shows that mental load can increase stuttering rate and that there is a subgroup of persons who stutter whose speech is highly sensitive to cognitive influences. Bosshardt and Fransen (1996) experimentally demonstrated that adult persons who stutter in comparison to persons who do not stutter need more time to process semantic information. With this background, it can be assumed that during speech production, people who stutter need more cognitive processing capacity and that these enhanced processing demands can interfere with the fluent execution of speech. In this view stuttering rate simultaneously depends on the extent of processing demands imposed by speech planning and on the individual sensitivity of persons to the interference that cognitive processes produce in concurrently executed speech movements. But this theoretical framework, which corresponds to the “demands and capacities model” of stuttering as proposed by Starkweather (1987)(p.75ff) can account for only part of the results of the present experiment. The problem for this explanation is the observation that in our experiment the two groups of persons were not significantly different with respect to the amount of stuttering induced by the secondary task. But in agreement with the expectations that can be derived from the demands and capacity framework it was found that a subgroup of stutterers was more sensitive to this kind of interference than most persons who do not stutter. On the other hand, it is also conceivable that the comparably smaller effect that mental load had on speech fluency of the low-interference group may be a result of the adaptation effect. Conclusions In sum, these data indicate that attention-demanding cognitive processes significantly influence stuttering, inhalation, and the timing of word repetitions. It is proposed that mental load reduces the speech of linguistic encoding and at the same time increases its defectiveness. The data suggest that people differ in the susceptibility of their speech to cognitive load and that cognitive load is particularly detrimental to the speech of a subgroup of individuals who stutter. Acknowledgements The author gratefully acknowledges that this research was supported by the Deutsche Forschungsgemeinschaft (DFG), grant no. BO 827/5-1. I am very grateful to Heiko Hübner who implemented the technical basis of this experiment, to Thorsten Zirkwitz who helped run the experiment, and to Serena L’hoest and Daniel Dobbert who scrupulously and very competently performed the acoustical analyses. In addition, I wish to tank Dr. Waltraud Ballmer and Dipl. Psych. Andreas Henning for their helpful comments with this project and manuscript. Finally, I gratefully acknowledge the help of Donald Goodwin with the preparation of the English version of this manuscript. References Anderson 1983 J.R. Anderson The architecture of cognitionHarvard University Press, Cambridge, Mass (1983) Andrews, Craig, Feyer, Hoddinott, & Neilson 1983 G. Andrews, A. Craig, A.-M. Feyer, P. Hoddinott, M. Neilson Stuttering: A review of research findings and theories circa 1982 Journal of Speech and Hearing Disorders, 48 (1983), pp. 226–246 Arends et al. 1988 N. Arends, D.-J. Povel, H. Kolk Stuttering as an attentional phenomenon Journal of Fluency Disorders, 13 (1988), pp. 141–151 Baddeley 1997 A. Baddeley Human memory (rev. ed.)Allyn and Bacon, Boston (1997) Baddeley, Eldridge, & Lewis 1981 A. Baddeley, M. Eldridge, V. Lewis The role of subvocalisation in reading Quarterly Journal of Experimental Psychology, 33A (1981), pp. 439–454 Baddeley, Grant, Wright, & Thomson 1975 Baddeley, A., Grant, S., Wright, E., & Thomson, N. (1975). Imagery and visual working memory. In P.M.A. Rabbitt & S. Dornic (Eds.). Attention and performance. London: Academic Press. Bernstein Ratner 1997 Bernstein Ratner, N. (1997). Stuttering. A psycholinguistic perspective. In R. Curlee & G. Siegel (Eds.). Nature and treatment of stuttering: New directions. Boston: Allyn & Bacon. Bosshardt 1997 Bosshardt, H.-G. (1997). Mental effort and speech fluency. In W. Hulstijn, H.F.M. Peters, & P.H.H.M. van Lieshout (Eds.), Speech production: motor control, brain research and fluency disorders. Amsterdam: Elsevier. Bosshardt and Fransen 1996 H.-G. Bosshardt, H. Fransen On-line sentence processing in adults who stutter and who do not stutter Journal of Speech, Language and Hearing Research, 39 (1996), pp. 785–797 Brutten & Trotter 1985 G.J. Brutten, A.C. Trotter Hemispheric interference: A dual-task investigation of youngsters who stutter Journal of Fluency Disorders, 10 (1985), pp. 77–85 Brutten & Trotter 1986 G.J. Brutten, A.C. Trotter A dual-task investigation of young stutterers and nonstutterers Journal of Fluency Disorders, 11 (1986), pp. 275–284 Ferreira 1991 F. Ferreira Effects of length and syntactic complexity on initiation times for prepared utterances Journal of Memory and Language, 30 (1991), pp. 210–233 Ferreira 1993 F. Ferreira Creation of prosody during sentence production Psychological Review, 100 (1993), pp. 233–253 Ford & Holmes 1978 M. Ford, V.M. Holmes Planning units and syntax in sentence production Cognition, 6 (1978), pp. 35–53 Gathercole & Baddeley 1993 S.E. Gathercole, A.D. Baddeley Working memory and languageLawrence Erlbaum, Hillsdale (1993) Glass & Mackey 1988 L. Glass, M.C. Mackey From clocks to chaosPrinceton University Press, Princeton, N.J (1988) Hitch 1978 G.J. Hitch The role of short-term working memory in mental arithmetic Cognitive Psychology, 10 (1978), pp. 302–323 Kamhi and McOsker 1982 A.G. Kamhi, T.G. McOsker Attention and stuttering: Do stutterers think too much about speech? Journal of Fluency Disorders, 7 (1982), pp. 309–321 Lindblom 1996 B. Lindblom Role of articulation in speech perception: Clues from production The Journal of the Acoustical Society of America, 99 (3) (1996), pp. 1683–1692 Lively et al. 1993 S.E. Lively, D.B. Pisoni, W. Van Summers, R.H. Bernacki Effects of cognitive workload and perceptual consequences Journal of the Acoustical Society of America, 93 (5) (1993), pp. 2962–2973 Logan & Conture 1995 K.J. Logan, E.G. Conture Length, grammatical complexity, and rate differences in stuttered and fluent conversational utterances produced by children who stutter Journal of Fluency Disorders, 20 (1995), pp. 35–62 Logan & Conture 1997 K.J. Logan, E.G. Conture Selected temporal, grammatical, and phonological characteristics of conversational utterances produced by children who stutter Journal of Speech and Hearing Research, 40 (1997), pp. 107–120 Mallard & Webb 1980 A.R. Mallard, W.G. Webb The effects of auditory and visual “distractors” on the frequency of stuttering Journal of Communication Disorders, 13 (1980), pp. 207–212 Max, Caruso, & Vandevenne 1997 L. Max, A.J. Caruso, A. Vandevenne Decreased stuttering frequency during repeated readings: A motor learning perspective Journal of Fluency Disorders, 22 (1997), pp. 17–33 Peters and Guitar 1991 T.J. Peters, B. Guitar Stuttering. An integrated approach to its nature and treatmentWilliams & Wilkins, Baltimore (1991) Power 1985 M.J. Power Sentence production and working memory The Quarterly Journal of Experimental Psychology, 37A (1985), pp. 367–385 Scott 1955 W.A. Scott Reliability of content analysis: The case of nominal scale coding Public Opinion Quarterly, 19 (1955), pp. 321–325 Starkweather 1987 C.W. Starkweather Fluency and stutteringPrentice-Hall, Englewood Cliffs, N.J (1987) St. Louis 1979 St. Louis, K. (1979). Linguistic and motor aspects of stuttering. In N.J. Lass (Ed.), Speech and language. New York: Academic Press. Sussman 1982 H.M. Sussman Contractive patterns of intrahemispheric interference to verbal and spatial concurrent tasks in right-handed, left-handed and stuttering populations Neuropsychologia, 20 (1982), pp. 675–684 Thompson 1985 A.H. Thompson A test of the distracting explanation of dysfluency modification in stuttering Journal of Fluency Disorders, 10 (1985), pp. 35–50 van Lieshout, Hulstijn, and Peters 1996a P.H.H.M. van Lieshout, W. Hulstijn, F.M. Peters Speech production in people who stutter: Testing the motor plan assembly hypothesis Journal of Speech and Hearing Research, 39 (1996), pp. 76–92 a van Lieshout, Hulstijn, and Peters 1996b P.H.H.M. van Lieshout, W. Hulstijn, F.M. Peters From planning to articulation in speech production: What differentiates a person who stutters from a person who does not stutter? Journal of Speech and Hearing Research, 39 (1996), pp. 546–564 b Wickens 1984 C.D. Wickens Engineering psychology and human performanceCharles E. Merrill Publishing Company, Columbus (1984) Wieland-Eckelmann 1992 R. Wieland-Eckelmann Kognition, Emotion und psychische Beanspruchung (Cognition, emotion and mental load)Hogrefe Verlag, Göttingen (1992) Yaruss 1997 J.S. Yaruss Utterance timing in childhood stuttering Journal of Fluency Disorders, 22 (1997), pp. 263–286 Address correspondence to Hans-Georg Bosshardt, Department of Psychology, Ruhr-Universität Bochum, P.O. Box 102148, D-44780 Bochum, Germany. 1 Preliminary analyses with the reduced data set (n = 15) showed that the factor session was significant but revealed no significant interaction of this factor with any of the other design factors. 2 Participant S19 is one of the three persons whose data were excluded from the following analyses because her mental calculation performance was not comparable to that of the persons who do not stutter; however, this participant’s data were used for the determination of reliability of stuttering rate. 3 In order to obtain more fine-grained information about the sound durations within a word, the following segmental durations for one word “Traubensaft” (grapejuice) were measured: a) the transition between the initial consonant /t/ and the beginning of the vowel (CV1); b) duration of the nucleus of the first vowel (Vow1); c) the time interval between the first and third vowel (VV13); d) the duration of the third vowel (Vow3); and e) the final transition between the end of the third vowel and the following consonant (VC3). The factors task, session, and repetition did not significantly interact with segments and no higher-order interactions of any of these factors with segment was significant. Only the group factor significantly interacted with segments (F(4,68) = 9.61; MSE = 4028.2; ϵ = 0.71; p < .000). This interaction was largely due to the fact that only the duration of CV1 was not significantly longer for persons who stutter than for persons who do not stutter. In comparison to persons who do not stutter, persons who stutter prolonged the sound transitions except the consonant-vowel transition in the first stressed syllable. These results suggest the factor task affects segmental durations in a way that is comparable to that of total word duration. Therefore, segmental durations were not analyzed in greater detail. 4 Alternatively, it can be proposed that the low-interference subgroup relies more on the visual sketch-pad for mental calculation than on central executive. Consequently, when speaking under dual-task conditions this subgroup is more fluent because word repetition and mental calculation largely draw on different resources. However, to my knowledge there is no other independent evidence which supports the assumption that there are subgroups of persons who stutter which differ with respect to their ability to use the visual sketch-pad. The empirical results of the present experiment do not permit a decision between this and the alternative interpretations. Presently, I prefer more cautious interpretation based on the adaptation assumption which is most directly related to the experimental procedure used here. Copyright © 1999 Elsevier Science Inc. All rights reserved. Recommended articles Effects of concurrent cognitive processing on the fluency of word repetition: comparison between persons who do and do not stutter 2002, Journal of Fluency Disorders more Working memory involvement in stuttering: Exploring the evidence and research implications 2007, Journal of Fluency Disorders more The effect of fast speech rate on stuttering frequency during delayed auditory feedback 2002, Journal of Fluency Disorders more View more articles » Citing articles (26) Related book content